Why are you contorting yourself into just one query? Your last clause would be a complete table scan seems to me. Sounds slow versus 2 queries.
Why can't you just do this? (pseudo code here) -- no table scans involved at all. select docid from b where t match 'blah'; if (rowcount > 0) // the we have a b so check for a mydocid=result select docid from a where docid=mydocid; if (rowcount > 0) // then a has the same docid mydocid=result end else // no b exists so a is the only potential one select docid from a where t match 'blah'; mydocid=result end Michael D. Black Senior Scientist Advanced Analytics Directorate Advanced GEOINT Solutions Operating Unit Northrop Grumman Information Systems _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users